Hyperspectral Remote Sensing in Lithological Mapping, Mineral Exploration, and Environmental Geology: An Updated Review

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/353245364
Hyperspectral remote sensing in lithological mapping, mineral exploration, and

environmental geology: An updated review
Article in Journal of Applied Remote Sensing · July 2021

DOI: 10.1117/1.JRS.15.031501
CITATIONS READS
128 2,800
2 authors:
Sima Peighambari Yun Zhang

Payame Noor University University of New Brunswick
11 PUBLICATIONS 253 CITATIONS 85 PUBLICATIONS 1,728 CITATIONS
SEE PROFILE SEE PROFILE
All content following this page was uploaded by Sima Peighambari on 13 April 2022.
The user has requested enhancement of the downloaded file.

REVIEW
Hyperspectral remote sensing in lithological mapping,
mineral exploration, and environmental geology:
an updated review
Sima Peyghambari* and Yun Zhang

University of New Brunswick, Geodesy and Geomatic Engineering Department, Fredericton,
New Brunswick, Canada
Abstract. Hyperspectral imaging has been used in a variety of geological applications since
its advent in the 1970s. In the last few decades, different techniques have been developed by
geologists to analyze hyperspectral data in order to quantitatively extract geological information
from the high-spectral-resolution remote sensing images. We attempt to review and update vari-
ous steps of the techniques used in geological information extraction, such as lithological and
mineralogical mapping, ore exploration, and environmental geology. The steps include atmos-
pheric correction, dimensionality processing, endmember extraction, and image classification.
It is identified that per-pixel and subpixel image classifiers can generate accurate alteration
mineral maps. Producing geological maps of different surface materials including minerals and
rocks is one of the most important geological applications. The hyperspectral images classifi-
cation methods demonstrate the potential for being used as a main tool in the mining industry and
environmental geology. To exemplify the potential, we also include a few case studies of differ-
ent geological applications. © 2021 Society of Photo-Optical Instrumentation Engineers (SPIE) [DOI:
10.1117/1.JRS.15.031501]
Keywords: hyperspectral remote sensing; geological applications; mineral spectra; lithological
mapping; ore exploration; environmental geology.
Paper 210082V received Mar. 10, 2021; accepted for publication Jun. 25, 2021; published online
Jul. 14, 2021.
1 Introduction
High-spectral resolution (hyperspectral) remote sensing has been used for Earth observation
since the advent of imaging spectrometer systems.1 Hyperspectral sensors can acquire images
in 100 to 200 contiguous spectral bands, to provide a unique combination of spatially and
spectrally contiguous images.2 Thanks to its ability to capture unique spectral signatures of the
surface material, hyperspectral imaging has been used in various Earth observation applications
including earth science, forestry, geography, agriculture, hydrography, atmosphere, climate
change monitoring, military, security and law enforcement.3
In earth science, geologic remote scientists have also utilized the advantages of hyperspectral
imaging in different geological applications, such as mineral industry, water quality determina-
tion, oil, and gas industries. They conducted hyperspectral imaging at various scales from close
range imaging including rock samples,4 cores,5 and outcrop scanning6 to airborne and space-
borne acquisition.7,8
There are several review papers on the topic of geologic hyperspectral remote sensing,
including applications of hyperspectral remote sensing in geology,9 multi- and hyperspectral
geologic remote sensing,7 mineral mapping using hyperspectral data,10 spectral processing meth-
ods for geological remote sensing,11 hyperspectral remote sensing and geological applications,12
and hyperspectral remote sensing for mineral exploration.13
However, an up-to-date review that covers the various components of using hyperspectral
remote sensing for geological applications is needed for future research. This paper attempts
to fill the gap by providing an updated review on hyperspectral missions, spectral properties
*Address all correspondence to Sima Peyghambari, sima.peyghambari@unb.ca
Journal of Applied Remote Sensing 031501-1 Jul–Sep 2021 • Vol. 15(3)
Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Peyghambari and Zhang: Hyperspectral remote sensing in lithological mapping, mineral exploration. . .
of diagnostic minerals, and various techniques for geologic information extraction from space-
borne/airborne hyperspectral images. The techniques include preprocessing, dimension reduc-
tion, endmember retrieval, and important image classification methods.
2 Hyperspectral Sensors
Over the last decades, airborne hyperspectral sensors including AVIRIS, HYDICE, DAIS, and
HyMAP overwhelmed the hyperspectral imaging. However, using spaceborne hyperspectral
sensors makes this technology more available for researchers. The launch of NASA’s EO-1
Hyperion sensor with 242 spectral bands in 0.4 to 2.5 μm made the new beginning in hyper-
spectral remote sensing. Several further spaceborne hyperspectral sensors have also launched in
the target of scientific research and land imaging, Earth observation, and natural resources and
atmosphere such as Tiangong-1, EnMAP, and PRISMA. However, there are plans for future
development of hyperspectral satellite sensors including HypXIM and HyspIRI. Table 1 repre-
sents some of the main hyperspectral missions with the target of Earth observation. Although
most of the spaceborne hyperspectral imageries (HSIs) have moderate spatial resolution (30 m)
(Table 1), some of the hyperspectral sensors such as PRISMA and forthcoming HypXIM use the
advantages of panchromatic sensors with the resolution of 5 and 2 m, respectively.
3 Spectral Properties of Different Minerals

Hyperspectral imaging data in the 0.4- to 2.5-μm visible- and near-infrared (VNIR) and short-
wave infrared (SWIR) spectral range are diagnostic in mineralogic and lithological mapping in
different climate and tectonic settings.12 Due to distinct spectral absorption feature, many min-
erals and rocks are distinguishable through their spectral patterns.9,14,15 Figure 1 shows the accu-
racy of different minerals in various wavelengths including VNIR, SWIR, and long-wave
infrared (LWIR) or thermal infrared (TIR).16 Therefore, hydrothermally altered and unaltered
rocks can be discriminated and mapped via the diagnostic signatures in the VNIR and SWIR
regions.17
Iron oxides and hydroxides including hematite, jarosite, limonite, and goethite show distinct
spectral absorption in VNIR from 0.4 to 1.1 μm. Iron oxides and hydroxide minerals are mostly
developed during the hydrothermal alteration processes associated with different ore bodies such
as porphyry copper deposits (PCDs).17
Hydroxyl-bearing minerals such as phyllosilicate, clay minerals, sulfate, and carbonate
groups show distinct spectral absorption in SWIR radiation region. Different minerals including
Al-Si-(OH) and Mg-Si-(OH)-bearing minerals such as clay minerals, talc, and chlorite, and
Ca-Al-Si-(OH)-bearing sorosilicate minerals such as epidote and OH-bearing sulfates such as
alunite and gypsum, and carbonates have diagnostic spectral characteristics in SWIR region.
Due to internal structures of the quartz mineral, it shows absorption signature around 8.3 to 9.1,
which belongs to TIR region.18
Spectral signatures of some hydrothermal minerals are presented in Figs. 2(a) and 2(b).
Figure 2(a) reflects the absorption features of the Al-O-H, Mg-O-H minerals, alunite, and car-
bonates in SWIR region, whereas Figure 2(b) shows the spectra features of the Fe oxides and
hydroxides.19 These minerals mostly associated with different hydrothermal alterations around
various ore deposits. Therefore, spectral feature of the minerals, which is one of the most impor-
tant aspects of hyperspectral remote sensing, can be used for lithologic and mineral mapping
as well as ore exploration. Figure 3 shows the spatial distribution of different minerals with
diagnostic spectral properties from VNIR and SWIR AVIRIS hyperspectral data over Cuprite,
Nevada.
There are several well-characterized spectral repositories of minerals and rocks such as
United States Geological Survey (USGS), John Hopkins University (JHU) and Jet Propulsion
Laboratory (JPL), Emission and Reflectance Spectral Library (EARSL), and National Mineral
Collection (NMC) spectral library. The USGS provided spectral library of minerals and rocks
from 0.35 to 2.5 μm regions. The JHU spectral library covers from 2 to 25 μm regions.20 NASA
JPL spectral library includes minerals with spectral features from 0.4 to 2.5 μm. The Council of

Table 1 The specific spectral characteristics of various hyperspectral missions.
Satellites/ Launch Number of Spectral Spectral GSD Swath Spectral imaging

No. Sensor platform time Organizations bands range (μm) resolution (nm) (m) width (km) techniques Application
1 Hyperion EO-1 2000 NASA 19 to 62 0.4 to 2.5 10 30 7.7 Grating and pushbroom Earth observation; mining, geology,
forestry, land, and area mapping

2 CHRIS PROBA 2001 ESA 19 to 62 0.4 to 1.0 1.25 to 11 25 to 50 13 Prism, pushbroom, Land and water related
and multi-viewing applications and aerosol
measurements
3 MERIS ENVISAT 2001 ESA 520 0.39 to 1.04 1.25 300 1150 Grating, pushbroom, Ocean observation
Journal of Applied Remote Sensing

(transmit) and onboard band
width selection
4 FTHSI HJ-1A 2008 NCDR/SEPA 115 0.45 to 0.95 4 100 50 Fourier interferometer Earth observation
5 TianGong-1 Shenzhou-8 2011 Chinese Academy of 128 0.40 to 0.25 10 VNIR 30 10 Pushbroom Scientific research and
Science and Physics 23 SWIR land imaging
031501-3
6 AHSI Gaofen-5 2018 Shanghai Institute 330 0.40 to 2.50 5 VNIR 30 60 Grating and pushbroom Earth observation
10 SWIR
7 HysIS IMS-2 2018 COSPAR ID 256 0.40 to 2.40 10 30 30 Dispersive (?) Agriculture, forestry,

and geography
8 PRISMA PRISMA 2019 ASI 237 0.40 to 2.51 12 30 30 Prism and pushbroom Natural resources
and atmosphere
9 HISUI ALOS-3 2019 METI 185 0.40 to 2.50 10 VNIR 30 30 Grating and pushbroom Energy and vegetation
12.5 SWIR monitoring
10 HyspIRI HyspIRI 2020 JPL, NASA >200 0.38 to 0.25 10 60 145 Pushbroom Volcanic, vegetation, soil,
and exploration
Jul–Sep 2021
•
11 EnMAP German HS 2021 GFZ/DLR 244 0.42 to 2.50 5 and 10 30 30 Prism and pushbroom Earth observation
12 HypXIM HypXIM 2023 CNES 214 0.4 to 0.25 10 8 15 Pushbroom Soil, urban, and coastal
biodiversity
Vol. 15(3)
Fig. 1 Qualitative accuracy of spectral analyses in VNIR, SWIR, and LWIR ranges.16
Fig. 2 Laboratory spectral features of different hydrothermal minerals including (a) clay minerals
and (b) iron oxides.18
Industrial Research Organization of Australia has developed the EARSL of different minerals.
The NMC produced by Geological Survey of Canada contains more than 100,000 minerals.21
Although spectral libraries of different minerals produced in laboratories have been widely
used in mineral and rock identifications, some natural heterogenies existed in various minerals
and associated spectral mixing possibly still have unwanted effects on the image processing
results.22,23 Field spectroscopy is evolving as a robust technique to overcome this issue specially

Fig. 3 Mapping results from the analysis of (a) VNIR and (b) SWIR AVIRIS hyperspectral data
from Cuprite area, Nevada.15
in mineral mapping and ore exploration. Hyperspectral data from cores provide information
about different mineral grades in ore exploration.10 Combination of field spectroscopy with
spaceborne/airborne hyperspectral images can get reliable results from image processing.
4 Hyperspectral Data Processing
4.1 Hyperspectral Image Preprocessing

Different levels of preprocessing are available to correct the original hyperspectral images
including geometric and atmospheric correction (ATCOR). Bad characteristics of an image that
have been caused by the sensor can be removed by geometric or sensor error correction tech-
niques. These errors involve stripes and other noise type distortions.24,25 For example, several
vertical lines with no information can be produced by detectors in Hyperion sensor. The infor-
mation from these areas can be covered by replacing their pixel values with the average pixel
values of their neighboring pixels.25
Due to scattering and absorption of solar radiation and spectra by atmospheric gases and
aerosols, the hyperspectral images have atmospheric effects. These atmospheric effects must
be removed in order to use hyperspectral data for quantitative remote sensing.26 ATCOR meth-
ods have evolved through the years so that they can be categorized as scene-based empirical
approaches and radiative transfer model (RTM)-based approaches.25–28 Several scene-based
empirical approaches were developed to correct the atmospheric effects on the hyperspectral
images. Kruse23 computed the average spectrum of a scene by internal average spectrum of
a scene. Then the corresponding spectrum of any pixel is divided by the average spectrum
to get relative reflectance spectrum for each pixel. This method was mostly applied for areas
without any vegetation. Roberts et al.29 introduced flat field correction approach. They normal-
ized the input image to an assumed area with neutral spectrally reflectance (with flat topographic
and spectral feature). The empirical line approach30 uses field reflectance spectra for bright and
dark targets to linearly correlate the raw input imaging spectrometer data. Reinersman et al.31
developed empirical “cloud shadow” method for ATCOR over the darker water surfaces. The
quick ATCOR approach32 is developed to remove atmospheric effects from multispectral and
hyperspectral images in VNIR and SWIR bands. This approach drives atmospheric compensa-
tion factor directly from imaging spectrometer data.

RTMs are physically based codes that try to simulate the transfer process of an electromagnetic
wave in the atmosphere.33 They need field-based measurement of the atmosphere conditions at the
time of image attainment and produce apparent reflectance or scaled surface reflectance.25,34 There
are several RTM-based algorithms to model the ATCOR. Atmospheric removal algorithm was
developed by Gao et al.35 This model performed to retrieve ground reflectance spectra from hyper-
spectral data by a theoretical modeling technique. It simulates the absorption and scattering effects
of gases and aerosols existing in the atmosphere. ATCOR developed by Richter36 includes a large
database of ATCOR functions. It covers a wide range of atmospheric conditions. Kruse27 devel-
oped ATCOR now algorithm, which is model-based ATCOR software using MODTRAN-4 code
to reduce atmospheric and topographic effects in the data.
Fast line-of-sight atmospheric analysis of spectral hypercubes was developed by Adler-Golden
et al.37 It retrieves the land surface reflectance without using ground measurements. Staenz et al.38
developed imaging spectrometer data analysis system to process hyperspectral data by removing
sensor and calibration artifacts. It converts at sensor radiance to surface reflectance and analyses the
hyperspectral data. Qu et al.39 developed high-accuracy atmospheric correction for hyperspectral
data. This model corrects the water vapor and other gases (such as CO2 and methane).
4.2 Dimensionality Processing

Once corrected, the hyperspectral data still have redundant spectral information, which need to
be processed. To reduce the computational costs and processing time without losing the useful
information, the data should be dimensionally diminished. Dimension reduction techniques can
be categorized into different groups such as unsupervised and supervised, linear and non-linear
approaches. Principal component analysis (PCA), minimum noise fraction (MNF), and indepen-
dent component analysis (ICA) are the most popular unsupervised dimension reduction
approaches. PCA is a linear widely used technique that searches to magnify the variance within
the new and lower subspace.40 MNF is a well-known denoising technique that transforms noisy
original hyperspectral data cube to channel images with the increasing noise level.41 ICA is a
statistical mechanism that transforms the data into maximally independent and non-Gaussian
subparts.42,43 This method uses virtual dimensionality to determine the number of independent
components, which should be retained.25 Supervised methods are performed to extract more
useful information using the prior knowledge of training samples. Commonly used supervised
dimension reduction algorithms include Fisher’s linear discriminant analysis44 and discrimina-
tive locality alignment (DLA).45 Local Fisher’s discriminant analysis46 and semisupervised
DLA45 are some variants of those techniques.
In order to overcome some problems that arose from traditional dimension reduction meth-
ods, sparsity-based techniques such as sparse-graph-based discriminant analysis47 and multiple-
features-combining48 method were developed to enhance the dimension reduction results. The
incorporates the spectral, texture apart from traditional methods; new methods based on machine
learning communities have been developed for processing hyperspectral images. A number of
metric learning algorithms such as relevant component analysis,49 neighborhood component
analysis,50 information-theoretic metric learning,51 and ensemble discriminative local metric
learning52 solve the problems in hyperspectral analysis. In addition, different manifold learn-
ing-based dimension reduction methods are proposed to display the non-linear structure in
hyperspectral images including locally linear embedding,53 Laplacian eigenmap,54 isometric
mapping,55 neighborhood preserving embedding (NPE).56 Hypergraph learning methods have
also been introduced to explore the multiple adjacency relationship in hyperspectral data and
discover the complex geometric structure between hyperspectral images.57 Discriminant
hyper-Laplacian projection,57 semisupervise hypergraph embedding,58 local pixel NPE,59 and
spatial–spectral regularized sparse hypergraph embedding57 are some types of graph embedding
methods for dimensionality reduction of hyperspectral images.
4.3 Endmember Extraction Methods

Due to enhanced spectral resolution of HSI, extraction of hyperspectral endmember is a funda-
mental step in the hyperspectral data processing. An endmember is a pure and diagnostic

signature that can be used to specify a spectral class. It should be noticed that each endmember
does not need to be a pure pixel.60 There are several endmember extraction algorithms that are
broadly grouped into two main classes including convex geometric-based and statistical-based
categories.12 In widely used linear unmixing model, the spectra of each pixel vector are regarded
to be a linear combination of endmembers weighted by their corresponding abundances.61,62
There are two main groups of endmember extraction methods within the linear spectral unmix-
ing algorithms as pure-pixel and non-pure-pixel methods. In these methods, the hyperspectral
data are considered as simplex, and its endmembers are constituting its vertices. The pure-pixel
approach assumes that there is at least one endmember in the image scene and applies simplex
growing algorithms or maximum volume transform to find the endmembers within the hyper-
spectral data cloud.63,64 The popular algorithms involve pixel purity index (PPI), and N-finder
algorithm (N-FINDR), vertex component analysis, automatic target generation process, and con-
vex cone analysis are typical pure-pixel or simplex growing methods.47,48 In order to overcome
the issues in pure-pixel endmember extraction, non-pure methods have been developed that use
the shrinking simplex algorithms or minimum volume transform techniques.64 Optical real-time
adaptive spectral identification system (ORASIS), minimum volume constrained non-negative
matrix factorization (MVC-NMF), minimum volume enclosing simplex (MVES), minimum vol-
ume simplex analysis (MVSA), and abundance separation and smoothness constrained non-neg-
ative matrix factorization are developed as non-pure-pixel extraction algorithms.64 The ORASIS
as the linear mixture model, developed by the Naval Research Laboratory, consists of series of
stepwise algorithms including prescreener, basis selection, and endmember selection to unmix
each pixel of dataset to different endmembers.65,66 This method considers all the data pixel inside
a simplex with endmember vertices. According to Chang,66 the functionality of this method
declined in hyperspectral data with low signal-to-noise ratio (SNR). MVC-NMF has been devel-
oped by Miao and Qi67 as a non-pure pixel-based algorithm, which is convex geometry-based
algorithm. This method tries to fit simplex while its volume is minimal and encompass the data
cloud.67 Nouri et al.68 used particle swarm optimization to improve MVC-NMF method for min-
eralogical unmixing of hyperspectral data with high SNR.68 MVSA is another linear mixture
model for endmember extraction that was developed by Li and Bioucas-Dias.69 This algorithm
tries to fit a minimum volume simplex to the hyperspectral data, which contains the abundance
fractions to belong to the probability simplex.70 In this method, it is assumed that pure pixel may
not exist in the hyperspectral data, which can address a common situation in hyperspectral data
with highly mixed pixels.70 Motivated by Craig’s belief, Chan et al.71 developed linear-based
model called MVES algorithm in which vertices of the simplex enclose all the observed pixels.
In this method, the simplex should estimate all the endmember signatures with high fidelity.72
Non-linear geometrical unmixing methods have been used less than linear techniques.
Broadwater73 introduced non-linear kernel methods to solve the unmixing problem in high-
dimensional hyperspectral data. Heylen74 introduced geodesic simplex volume maximation
methods under non-linear mixing assumptions to extract hyperspectral endmembers. The stat-
istical-based endmember extraction methods use statistical representations.63 Based on paramet-
ric, non-parametric, and spatial statistic representations, different statistical-based unmixing
methods have been developed. Stochastic mixing model, which assumes Gaussian distribution
for each endmember, is an example of statistic-based algorithms that used parametric
representations.75 ICA is proposed by non-parametric statistical unmixing approaches. ICA-
based abundance quantification algorithm and combination of ICA and independent factor
analysis are some kinds of non-parametric statistical endmember extraction approach.63
Automated morphological endmember extraction, iterative error analysis, and spatial–spectral
endmember extraction (SSEE) are statistical unmixing methods that used spatial statistics to
improve the endmember selection.12,63
5 Image Classification Techniques

Image classification generally refers to a set of techniques to assign different classes to all the
pixels in the digital image. Since the advent of hyperspectral remote sensing, different hyper-
spectral image classification approaches and algorithms have been developed and evolved.

Due to high dimensionality, mixed pixels, and a smaller number of training samples, image
classification procedure for high-spectral-resolution hyperspectral images is facing more chal-
lenges. However, various methods have been designed and employed to overcome these chal-
lenges. The image classification mainly categorized into three main groups including supervised,
unsupervised, and semisupervised techniques. Over the last decades, based on various criteria,
supervised and unsupervised image classification has been categorized into further groups
including per-pixel and subpixel, parametric and non-parametric, soft and hard, spectral and
spectral–spatial, and per-field.25,76,77
In supervised classification, the image analyst uses training samples (known) obtained from
expert knowledge to specify different spectral signatures or pixel values of the image as to differ-
ent classes.25 Based on prior knowledge, the user selects sample representative pixels of a known
cover type or pattern in an image as the specific class and assign it as training sites. Then these
training sites will be used as references for the classification of other pixels in the image. Some
of the important supervised classifiers include maximum likelihood classifier (MLC), support
vector machine (SVM), spectral angle mapper (SAM), decision tree, and artificial neural net-
work (ANN).25
In unsupervised classification techniques, there is no need for the analyst’s extensive prior
information and contribution. Helping to find different clusters in data, this approach can be
utilized for feature extraction and segmentation. In this method, the classes are created purely
based on spectral information not manual visual interpretation. K-means, iterative self-organ-
izing method (ISODATA), and clustering are some of the common unsupervised classifiers.78
Semisupervised classification utilizes some available known reference information along
with undefined data. Semisupervised random was forest recently used for hyperspectral image
classification.79
Parametric classification techniques consider data to have a normal distribution pattern and
deal with statistical parameters such as mean vector and covariance matrix. In the case of
complex landscape, the classification results should be noisy. In addition, insufficient, non-
representative, and multimode-distributed training samples result some ambiguities in image
classification. Combining spectral data with ancillary information and non-statistical data in
parametric approach make some difficulties for remote sensing image classification.77 However,
due to its robustness and easy implementation, the MLC is one of the most widely used para-
metric classification.
In non-parametric classifier, no statistical parameters of the distribution of the input data are
required. It is facilitated in this approach to integrate non-remote sensing data in the image clas-
sification procedure. Decision tree classifier, expert systems, SVM, deep learning, and neural
network become typical examples of non-parametric classifiers widely deployed in remote
sensing image classification.25
Per-pixel classifier creates and assigns a signature through combing the spectra of the feature
set from the entire pixel to a single class. Not considering the mixed pixel problem, the resultant
signature includes the integration of all the materials available in the training set of pixel.25,77
Most of the classifiers such as MLC, Euclidian distance, ANN, decision tree, and SVM, SAM,
binary encoding, and spectral feature fitting are important per-pixel classifiers. Due to poor
spatial resolution of hyperspectral remote sensing images, each pixel spectrally is not pure and
normally contains a mixture of two or more target materials.
Subpixel classifiers have been introduced to overcome the challenge of existing different
materials in a pixel. These techniques consider the spectral value of each pixel as the integration
result of linear or non-linear combination of pure materials. These techniques assign the exact
class for each pixel for the classification of the medium- to low-spatial-hyperspectral remote
sensing images. It employed different linear and non-linear unmixing models for subpixel level
spectral matching.25,76
Linear unmixing models consider each pixel as a linear combination of a set of spectral
fingerprints as endmembers respect to their abundances. In other words, in linear unmixing mod-
els, the subpixel component can be physically distinct in their reflectance property. In geological
applications, the different mineral fields or library spectra have been applied for mineral iden-
tification and their abundance. The most popular model for spectral unmixing is linear mixing

model (LMM). Some endmembers or pure pixel detection algorithms such as N-FINDR and PPI
are performed in this model.
Non-LMMs consider that the light photon has been affected by different earth surface mate-
rials before detection by the sensor. Various types of photon interaction in the mixed pixel in non-
LMM are vertical mixing (tree canopies) and horizontal mixing (granular mixtures) and both.
Bilinear models, intimate mixture models, neural-network-based models, kernel-based models,
physical modeling, and manifold techniques are the non-LMMs have been used by many
researchers.77,80
Spectral classifiers consider hyperspectral images as only spectral data without any concerns
about spatial information. Due to high level of mixing in the spatial distribution of the
different classes, spectral classifiers produce noisy results.25 Although the spectral classifiers
have the conceptual simplicity and computational effectiveness, they are not able to effectively
separate a number of land cover materials. MLC and ANN are some of the important spectral
classifiers.
Spatial-contextual approaches were developed to get higher accuracy classification.25 In
these techniques, spatial information from adjacent pixels are extracted to get a better classifi-
cation result. Some smoothing techniques including fuzzy logic and neural network can be per-
formed before the main classification approach. Texture extraction, MRFs modeling, image
segmentation, and object-based image analysis are the main spatial-contextual remote sensing
methods. Spectral–spatial classifiers are mostly considered for hyperspectral image processing.
In these approaches, parametric and non-parametric classifiers are performed before the employ-
ment of the spatial classifiers.76,81
Soft classifiers employ conditional probability to make a decision boundary for classification
of target in the image.77 In this method, each pixel can belong to more than one class and there is
a gradient between each class. Linear mixture modeling and Fuzzy classification are the most
common soft classification technique.
Hard classifiers in contrast make the decision boundary of the target without doing the prob-
ability assessment. They consider each pixel belonging to the class it most closely resembles.
The results of this classification have lower accuracy than soft classification approaches.
ISODATA, parallelepiped, K-means, maximum likelihood, and neural networks are the most
important hard classifiers.76
Considering single or multiple classifier(s) are another criterion to group image classifiers.25
Single classifier assumes a class label to a given pixel. One of the most well-known classifiers
that have been used as a single classifier is SVM. It creates linear decision boundaries to classify
multiple classes. This method defines different training samples by finding maximum margin
hyper planes in the space of the mapping sample.
In ensemble or multiple classifiers, a set of different classifiers are performed to assign a class
label for a given pixel. It could increase the accuracy of the classification result. In this approach,
it should be considered that combination of the classifiers has benefit to increase the accuracy
without any drawbacks.78,81
Prior reference data requirement also establishes a distinct category to group spectral image
classifiers. In the case of no prior reference data, the direct spectral patterns of a pixel are used to
image processing. In contrast, some techniques utilize the predefined representative information
as reference data for image processing procedure.11 These methods are categorized as knowl-
edge-based and data-driven approaches.
Knowledge-based techniques incorporate knowledge from spectral features of the different
materials. These approaches apply distinct characteristics of absorption features such as position,
depth, asymmetry, and width in different materials.11,14,19,82,83 Band calculation, feature mapping,
expert system, spectral deconvolution, wavelet analysis, and scattering theory constitute different
methods constituting knowledge-based techniques.
Data-driven methods require the hyperspectral data and additional reference data
(spectra).66,84 The reference data are commonly assigned as training classes or endmember
sets, which are imported from a spectral library or derived from the image. Different classi-
fications included in the per-pixel and subpixel classifiers are categorized in data-driven
classifiers.

6 Geological Applications
6.1 Lithological Mapping

High-spectral-resolution properties of the hyperspectral remote sensing make it as an important
potential to accurately recognize and map the earth surface materials.29,33 Hundreds of contigu-
ous spectral bands of hyperspectral images can be used to extract pixel spectrum for mineral and
rock discrimination.33 Performing mineral and lithological mapping from hyperspectral images
is generally due to comparison of unknown spectrum to a reference spectrum.82 Spectral absorp-
tion characteristics of different minerals are mostly related to vibrations of Fe2þ , Fe3þ , Al-OH,
Mg-OH, OH-, and CO3 in the VNIR and SWIR. Minerals with Fe2þ and Fe3þ have an absorp-
tion peak at 1.03 and 0.64 μm, respectively. Al-OH-bearing minerals have a significant absorp-
tion peak at 2.15 to 2.22 μm, whereas the most important absorption peak position of Mg-OH
minerals occur at 2.30 to 2.39 μm. CO3 group minerals show a distinct absorption peak at
2.35 μm.14,16,18–20 VNIR-SWIR region is mostly used for alteration minerals mapping including
Fe-OH, Mg-OH, Al-OH, and CO3 -bearing minerals. However, these spectral features can be
applicable for the Mg, Fe, and Al silicate minerals such as olivine, pyroxene, mica, and amphib-
ole. Mineral spectral features of silicate minerals mostly attributed to vibration of Si–O bonds in
TIR region.14,16,19
Kumar et al.85 proposed an automated lithological mapping approach on AVIRIS-NG
hyperspectral data from gold-bearing granite-greenstone belt of the Hutti area (India). In that
approach, they employed spectral enhancement techniques such as PCA and ICA and different
machine learning algorithms (MLAs) to get an accurate lithologic map (Fig. 4). However, they
utilized conventional geological map and spectral enhancement products derived from ASTER
data to generate a high-resolution reference lithology map. Among different MLAs, they found
SVM using joint mutual information maximization (JMIM)-based optimum bands to have
a better performance with higher accuracy.
Salehi et al.86 investigated the performance of HyMAP hyperspectral images and Sentinel-2,
ASTER and Landsat-8 OLI spaceborne multispectral data for lithological mapping of mafic–
ultramafic rocks in lichen-covered area in south west of Greenland. They performed EnMAP
geological mapper (EnGeoMAP) and iterative spectral mixture analysis (ISMA) algorithms.
They used structural similarity index measure to compare the output of classification results
of airborne data with available geological map and spaceborne data with reference HyMAP
data. According to their results, the HyMAP and spaceborne multispectral data can provide geo-
logical maps comparable to geologic maps over the less accessible arctic regions. Three different
ultramafic rocks (dunite, peridotite, and pyroxenite) and one mafic rock (gabbro) has been
mapped based on VNIR and SWIR spectral analyzing.86 They found EnGeoMAP algorithm
to have a better performance for dunitic rocks and ISMA method for peridotite and pyroxenite
units.
Harris et al.87 performed several methods on airborne PROBE hyperspectral data from the
test site in southern Baffin Island, Canada to produce lithological–compositional map. After
masking out and eliminating water bodies, ice, snow, and vegetation areas from analysis, they
applied MNF transformation to reduce the large dataset to a fewer number of components that
have the main information. They performed matched filtering (MF) to extract the user defined
endmembers due a partial unmixing method. The end members were selected by color discrimi-
nation in MNF composite images. They discriminated one lithological group (metatonalites) and
three compositional units (psammite, quartzite, and monzogranites).
Zhang and Li88 implemented an evolved SAM approach using EO-1 Hyperion hyperspectral
images in lithological mapping in two different arid areas in China. They employed two different
methods to improve the lithological mapping performance. In the first method, they used the
mean of spectral derivatives and mean of original spectral data in SAM to improve class sepa-
rability. In the second approach, multiple reference spectra were utilized to accommodate the
spectral variability. As the result, they mapped Carboniferous-Permian volcanic-sedimentary
rocks, Jurassic sedimentary rocks, granite, and quaternary sediments in Junggar area and
Cambrian, Ordovician, Silurian, Devonian sedimentary rocks including dolomite, shale, marl,
siltstone, quartzose sandstone, and sandstone, and quaternary sediments in Kalpin area.

Fig. 4 (a) Reference lithology map derived from spectral enhancement products using ASTER
and (b) lithological classification map generated from SVM using JMIM-based optimum bands of
AVIRIS-NG hyperspectral data from gold-bearing granite-greenstone belt of the Hutti area (India).85
Dadon et al.89 used EO-1 Hyperion hyperspectral data for stratigraphic and lithologic map-
ping. They performed SAM supervised classification after some preprocessing and endmember
extraction processes. They separated different plutonic (monzogranite, granite, and granodiorite),
volcanic (green tuffs and basaltic lava), and sedimentary (limestone, sandstone, conglomerate, and
dolomite shale) rocks form the Dana Geology National Park (south-west Jordan) after stratigraphic
and lithologic mapping process.
Pal et al.77 considered the VNIR and SWIR bands of Hyperion spaceborne hyperspectral data
and ASTER and Landsat 8 multispectral images. They implemented four-step technique for
lithological mapping of Udaipur area, Rajasthan, western India. They employed a hybrid clas-
sification method involving minimum distance, SAM, spectral information divergence, and
SVM to optimize lithological mapping. Pal et al.77 used this method to discriminate different
sedimentary and metamorphic lithoclasses including quartzite, phyllite, graywacke, dolomite,
mafic-metavolcanics, migmatite, graphitic metapelites, and quartzite-arkose-conglomerate in the
Udaipur area.
6.2 Mineral Mapping and Mineral Exploration

Mineral mapping and determination of surface composition for mineral exploration purposes is
one of the most important applications of hyperspectral remote sensing. There are different ore
deposits classifications based on different fundamental genetic processes: magmatic, hydrother-
mal (magmatic), sedimentary, metamorphic, and mechanical process. Different mineralization
systems are dominantly associated with different alteration associations and mapping surface
mineralogy can be considered as vectors to ore deposits.90 A schematic overview of

Fig. 5 Generalized alteration-mineralization zoning pattern for PCDs.91
intrusion-related mineral deposits and associated alteration patterns have been shown in Fig. 5.91
Figure 6 shows some diagnostic different alteration assemblages along with some of the main ore
deposits. Most of the alteration minerals have distinct spectral features in VNIR-SWIR and TIR
region.92
Magmatic hydrothermal deposits include several important ore deposit types such as por-
phyry, epithermal, skarns, intrusion-related, volcanogenic massive sulfide (VMS), iron oxide
copper gold (IOCG) uranium rare earth element (REE), alkaline complexes, and spreading cen-
ters sea-floor smokers.93 Hydrothermal intrusion-related ore deposits directly formed within or
with a distance with igneous intrusions. There are spatial alteration patterns and mineral assemb-
lages in these deposits. However, they have some genetic relationship with the fractures and
faults in the host rocks. They include alteration patterns such as feldspathic, sericitic, silicic,
greisen, calc-silicate, and/or advanced argillic assemblages.
Jintanzi gold province (China) is one of the intrusion-related gold deposits. This deposit is
mostly related to quartz veins within the granites and associated with different order faults. Using
spaceborne Tiangong-1 HSI and airborne short-wave infrared airborne spectrographic imager
(SASI) data, Liu et al.94 tried to produce alteration mineral maps via SAM algorithm. They
considered JPL and endmembers extracted by SSEE as the reference spectra. Using either set
of the reference spectra, they detected alteration muscovite, kaolinite, chlorite, epidote, calcite,
and dolomite as the hydrothermal minerals. Figure 7 shows the mineral maps derived from image
endmembers from SASI and Tiangong-1 hyperspectral data. According to their results and pre-
vious studies, the distribution of the alteration minerals has been distinctly controlled by the
gold-bearing quartz veins and faults.
Epithermal Au–Ag deposits have been considered as important gold and silver sources in the
world, which commonly distributes along the convergent plate margins.95 Extensive geological
studies have been done on different epithermal gold-silver deposits. However, several research-
ers have considered the application of remote sensing data in order to investigate and model the

Fig. 6 Common alteration minerals in different ore deposits.92
hydrothermal alteration mineral mapping in these deposits.96–99 Silisic, potassic, argillic, propy-
litic, and zeolitic are the most common hydrothermal alteration minerals in epithermal deposits.
Rodalquitar deposit (southern Spain) is one of the most popular epithermal gold and silver
deposits that have been extensively studied in geochemical aspect to understand the minerali-
zation process. Also numerous geological remote sensing studies have been conducted to gen-
erate mineralogic map of the area. van der Meer et al.97 performed wavelength mapper and
QuanTools to derive absorption feature from HyMAP hyperspectral images of Rodalquiar epi-
thermal deposit. They used absorption feature position as a key to find the mineral chemistry
variations such as Al–Mg versus OH, which can be interpreted in terms of fluid compositions
and temperature. Figure 8 displays the alteration mineral maps of Rodalquiar epithermal system
constructed via Quantools and wavelength mapper.97
PCDs constitute the largest source of copper and a major source of molybdenum, gold, and
silver in the world.100 These deposits commonly have diagnostic broad alteration patterns with
distinctive minerals. Typical alteration types in PCDs include potassic alteration in the core,
which is surrounded by sericitic, argillic, and propylitic zones.91 Subtle variations in the alter-
ation minerals could be considered as the high economic potential in PCDs. For example, zona-
tion of chemical variation of white mica in phyllite alteration is a proxy of chemistry of ore
forming fluid.16,101 Figure 9 shows the variation of white mica chemistry, which has different
SWIR wavelength characteristics in typical cross section of PCD.16 Although several spaceborne
and airborne hyperspectral data have been used for porphyry Cu exploration, there are a few
literatures that investigated them.102–104 Sarcheshmeh PCD (southern Iran) located in the
Orumieh-Dokhtar magmatic arc is one of the best sites for studying alteration related to
PCD. Using Hyperion data, Zadeh et al.104 (2014) identified characteristic alteration minerals
including biotite, muscovite, illite, kaolinite, goethite, hematite, jarosite, pyrophyllite, and chlo-
rite via subpixel mixture tuned matched filtering (MTMF) method (Fig. 10). Discriminating
minerals such as biotite and iron oxide is one of the most important evidence for PCD explora-
tion. Bishop et al.102 used SAM and MTMF techniques to discriminate and map the alteration
mineral assemblages from Hyperion data in the Pulang, PCD (Yunnan, China). They determined
argillic alteration, iron-oxide-, and sulphate-bearing minerals in the target deposit.

Fig. 7 Mineral maps derived from the use of image endmembers applied to (a) SASI data and
(b) Tiangong-1 HSI data from Jintanzi gold province (China).94
IOCG (±U-REE) deposits are a group of magmatic-hydrothermal deposits, which are struc-
turally controlled and commonly associated with different alteration zones. The alteration zones
of host rocks begin with sodic alteration, which changed to sodic-calcic and potassic toward the
main Fe-oxide mineralization.105 Deposits with more developed potassic alteration commonly
have intensive magnetite mineralization. Surficial rocks often show sericitic and silisic alteration.
Investigation of Corriveau et al.106 show that each alteration type in this deposit type has a unique
spectral signature. Potassic and sodic alterations are mostly distinguished by the location of
hydroxyl feature (OH-) near 2.200 μm. According to their results, airborne or spaceborne hyper-
spectral data have a potential capability to map exposed alteration maps around the IOCG depos-
its. Tappert et al.107 used the infrared reflectance spectra to discriminate the high-Al and low-Al
phengite as a potassic mineral identified at the Olympic Dam IOCG deposit (south Australia).
The phengite minerals in the heavily sericitized ore-bearing rocks have lower Si content, higher
Al content, and lower Mg content that the phengites formed in the weakly sericitized altered host
rocks. High-Al phengite has a spectral absorption feature at 2.206 μm, whereas low-Al type is
recognized by the absorption feature at 2.213 μm. Therefore, spectral absorption feature can be

Fig. 8 From top, mineral map constructed with Quantools and mineral map constructed with the
wavelength mapper using HyMAP data of Rodalquilar epithermal system (southern Spanish Gabo
de Gata volcanic area).97
Fig. 9 Typical cross section of PCD, showing generalized Al(OH) 2.200 μm absorption of different
white mica in phyllic alteration.16,101

Fig. 10 Final classification image map of alteration minerals at Sarcheshmeh deposit (southern
Iran) derived from MTMF algorithm. Bio, Mu, Il, Kao, Goe, Hem, Ja, Pyr, and Ch indicate biotite,
muscovite, illite, kaolinite, goethite, hematite, jarosite, pyrophyllite, and chlorite, respectively.104
representative of the sericitic alteration degree due hydrothermal alteration related ore forming
process. Laukamp et al.108 used HyMap airborne hyperspectral data to derive high-resolution
alteration mineral map of Mount Isa Inlier IOCG deposit (Australia). They employed hyperspec-
tral data as a tool for hydrothermal alteration detection and fluid pathway identification related to
Cu–Au IOCG deposits.
VMS deposits are kind of metal sulfide deposit, mostly copper and zinc, which formed near
the sea floor. Similar to other magmatic hydrothermal ore deposits, they have alteration patterns,
which have been formed via circulation of ore forming fluid through the host rocks.109 The
hydrothermal alteration patterns include potassic, advanced argillic, argillic, sericitic, chloritic,
and carbonate propylitic from the inner mineralized core to the peripheral area.110

There are different hyperspectral remote sensing studies on the several important worldwide
VMS deposits to characterize the hydrothermal systems. van Ruitenbeek et al.111 used HyMAP
hyperspectral data to evaluate white mica distribution pattern around Panorama VMS deposit
(western Australia). They used the advantages of different absorption features at 2.185 to
2.235 μm of Al-rich and Al-depleted white mica to map the alteration patterns and alteration
related fluid. Using AVIRIS-NG hyperspectral data, Samani et al.112 discriminated alteration
minerals including calcite, muscovite, and chlorite in the Ambaji-Deri area (northwestern
India). They performed different image processing approaches such as MNF, PPI, N-dimen-
sional visualization, and SAM classification. They matched the endmember spectra with the
USGS spectral library. Figure 11 shows the observed spatial distribution of different hydrother-
mal minerals in the Ambaji-Deri region. Laakso et al.113 used different scale hyperspectral data
such airborne, laboratory, and field methods for VMS mineral exploration in Canadian Arctic.
They took the advantages of the spectral Al-OH and Fe-OH absorption features in the SWIR
wavelength region of biotite and chlorite, which reflect the chemical compositional changes with
the distance to the mineral deposit. Fe-OH has an absorption feature at 2.254 μm in proximal
areas to the ore deposit while it changes to 2.251 μm in the distal areas. In addition, proximal
areas have the Al-OH absorption feature at 2.203 μm in contrast with the absorption feature at
2.201 μm in distal areas.
Fig. 11 Spatial distribution of the carbonate minerals at northeast of Khokhar Bili and alteration
minerals around Jharivav and Amblimal.112

Skarn deposit is another type of magmatic hydrothermal deposits, which have been formed
due to metamorphism and hydrothermal alteration of carbonate-bearing rocks. They are consid-
ered as the potential source of different ore minerals such as copper, tungsten, iron, molybde-
num, zinc, and gold. There is a zonation pattern in skarn deposits from garnet in the proximal to
pyroxene in distal areas and pyroxenoid such as wollastonite in the skarn and marble contact.114
However, individual skarns may show systematic compositional changes in these zonation
pattern. Thus detail mineralogy of skarns and their zonation can be useful in ore exploration
processes.
There are several hyperspectral remote sensing studies to detect alteration zones around skarn
deposits as a proxy to skarn type ore deposit exploration. Xu et al.115 used spaceborne Hyperion
and close-range field hyperspectral data from Dapingliang skarn copper deposit (China) to iden-
tify the mineral zones around the skarn deposit. They distinguished pixels related to skarn using
SAM from spectra overall shape and absorption bands spectral shape. Field data were applied to
directly identify alteration mineral instead of just surface materials, which probably do not have
any direct relationship with the ore mineralization. Tian et al.116 used the SWIR spectral analyses
to detect alteration minerals including sericite group minerals (montmorillonite, illite, and mus-
covite), kaolinite, and carbonate minerals (calcite, ankerite, and dolomite), with minor chlorite,
halloysite, and dickite around Jiguanzui Cu-Au (Eastern China) deposits. According to their
results, Fe-OH absorption feature of chlorite (2.241 to 2.263 μm) becomes shorter to the distal
areas. Also minimum Al-OH absorption feature of 2.209 μm can be a useful vectoring tool to the
ore deposit.
6.3 Environmental Geology

Environmental geology is one of the geological applications to solve environmental problems
derived from interaction of human with geologic environments like mineral resources. For in-
stance, the mining projects and abandoned mines are one of the main agents contributing to
cause serious ecological pollution. Vegetation, soil, water resources, and human can be threat-
ened by wastes that were left over including lead, zinc, cadmium, and some toxic minerals.117,118
These wastes with a widespread distribution have the high potential of pollution. As mentioned
in the foregoing sections, hyperspectral remote sensing has a powerful potential to detect
earth surface materials such as different minerals and elements. Therefore, it can be used to
characterize and monitor the mineralogy of mine waste surfaces to predict the potential source
of pollution including metal leaching and acidity. Since the application of hyperspectral remote
sensing in geology, different studies have been done to monitor pollution from various mining
areas.117,119–125
One of the processes that have been caused pollution in the mining area is pyrite
oxidation.123,125 This process may produce acidic water, which has gradually crystallized differ-
ent Fe-bearing secondary minerals in the way from the mining waste. These minerals include
copiapite,61 jarosite ½ðK; H3 O; NaÞFe3 ðSO4 Þ2 ðOHÞ6 , schwertmannite ½Fe8 O8 ðOHÞ6 SO4 , ferri-
hydrite ½Fe5 HO8 ; 4H2 O, goethite ½α-FeOðOHÞ, and hematite ½α-Fe2 O3 .123 These secondary
minerals are spectrally identifiable due to their diagnostic spectral reflectance signatures.
Swayze et al.123 used AVIRIS spectral data to evaluate mine waste in the California Gulch
Superfund Site. They used 0.4 to 2.5 μm spectral range of hyperspectral data to detect secondary
Fe-bearing minerals and create a mineral map to highlight the areas, which have the potential for
acidic drainage and predict the leaching and acid generation sites at Leadville mining district
(Fig. 12). They performed Tetracorder algorithm to produce the Fe-bearing minerals map.
Davies and Calvin126 have used airborne hyperspectral data to characterize acidic mine waste
at the Leviathan mine Superfund site in the eastern Sierra Nevada. They used SAM and MF
algorithms to map minerals related to acid mine drainage.
Mielke et al.125 applied spaceborne hyperspectral and multispectral data to monitor mine
waste mineralogy in south Africa. They derived iron feature depth (IFD) as a new index to detect
the extend of waste minerals distribution. This index is justified with primary and secondary
iron-bearing minerals. They performed material identification and characterization algorithm
on the EO-1 Hyperion data for mineral identification. According to their results, integration

Fig. 12 Spectral traverse and AVIRIS mineral maps overlaid on a high-spatial-resolution aerial
photograph of the Venir mine-waste pile. Mineral maps show the spectrally dominant Fe-bearing
secondary minerals.123
of hyperspectral (EnMAP) and multispectral (Sentinel-2) for mineral mapping and IFD mon-
itoring could be a promising proxy of mine waste.
7 Discussion and Conclusion

Although hyperspectral remote sensing data have been extensively used in different geologic
fields such as lithological mapping, mineral exploration, and environmental geology in the last
few decades, there are still several restrictions that limit the use of hyperspectral remote sensing
in geological applications.
First, the availability of hyperspectral data is still very limited, and the available data only
cover certain areas of the world. This has limited the scope of using hyperspectral remote sensing
for geological applications. Although different airborne hyperspectral sensors (e.g., AVIRIS,
HyMAP, HYDICE, and DAIS) and different spaceborne hyperspectral sensors (e.g., EO-1
Hyperion, EnMAP, HISUI, HyspIR, PRISMA, HYPXIM, MSMI, and HERO) have been devel-
oped, the number of hyperspectral sensors that are currently in use is very limited. Therefore,
limited hyperspectral data are available with limited ground coverages.
Second, the spatial resolution (ground sampling distance or GSD) of the available spaceborne
hyperspectral images, such as EO-1 Hyperion, is still limited. To improve the spatial resolution,
the new spaceborne hyperspectral mission PRISMA added a panchromatic sensor with 5-m spa-
tial resolution, and the forthcoming space-based HypXIM mission will add a 2-m panchromatic
sensor. In addition, some researchers have attempted to combine spaceborne hyperspectral data
with airborne hyperspectral images or with higher spatial resolution multispectral data (such as
ASTER and Sentinel-2) to overcome the GSD limitation of spaceborne hyperspectral images.
Third, the SNR in the spectral bands of spaceborne hyperspectral images that are useful for
geological applications is too low. For example, the SNR values of Hyperion spectral bands in
VNIR region vary from 140:1 to 190:1, whereas their values are even lower in SWIR, 96:1 in
1.225 μm, 36:1 around 2.125 μm.3 The most useful bands to detect minerals are around 2.0 to
2.4 μm (SWIR), whereas the bands around 0.43 to 0.90 μm are important for detecting ferric
iron. According to Kruse et al.,8 SNRs of at least 100:1 in the SWIR are required for mineral
detection. Therefore, to improve the SNR, new spaceborne hyperspectral sensors have achieved
a significant advancement. For example, the DESIS hyperspectral sensor has reached an SNR of
∼200∶1; but its GSD is still 30 m.127 And the forthcoming hyperspectral sensors (i.e., PRISMA,

EnMAP, HISUI, SHALOM, HyspIRI, and HypXIM) are expected to have even higher SNR
values in both the VNIR and the SWIR spectral ranges.3 In addition, adjacent channel overlap
and data redundancy are other issues about hyperspectral data.
According to the characteristics of hyperspectral data and the nature of geological informa-
tion, the data processing approaches may differ from other conventional methods. Therefore, it is
expected that special hyperspectral image processing techniques will be developed in the future
to meet the requirements of geological applications.
Acknowledgments
This study was supported by the funding from the Canadian Space Agency’s FAST (Flights and
Fieldwork for the Advancement of Science and Technology) research grant (19FANBA36)
received by the second author. We are grateful for the constructive comments provided by the
editor and anonymous reviewers.
References
1. M. J. Abrams et al., “Mapping of hydrothermal alteration in the Cuprite mining district,
Nevada, using aircraft scanner images for the spectral region 0.46 to 2.36 μm,” Geology
5(12), 713–8 (1977).
2. S.-E. Qian, Hyperspectral Satellites and System Design, CRC Press, Boca Raton
(2020).
3. J. Transon et al., “Survey of hyperspectral Earth observation applications from space in
the Sentinel-2 context,” Remote Sens. 10(2), 157 (2018).
4. B. Cooper et al., “Midinfrared spectral features of rocks and their powders,” J. Geophys.
Res. Planets 107(E4), 1–17 (2002).
5. D. Green and M. Schodlok, “Characterisation of carbonate minerals from hyperspectral
TIR scanning using features at 14 000 and 11 300 nm,” Aust. J. Earth Sci. 63(8),
951–957 (2016).
6. T. Kurz and S. Buckley, “A review of hyperspectral imaging in close range applications,”
ISPRS Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. XLI-B5, 865–870 (2016).
7. F. D. Van der Meer et al., “Multi-and hyperspectral geologic remote sensing: a review,”
Int. J. Appl. Earth Obs. Geoinf. 14(1), 112–128 (2012).
8. F. Kruse, J. Boardman, and J. Huntington, “Comparison of airborne hyperspectral data and
EO-1 hyperion for mineral mapping,” IEEE Trans. Geosci. Remote Sens. 41, 1388–1400
(2003).
9. E. A. Cloutis, “Review article hyperspectral geological remote sensing: evaluation of
analytical techniques,” Int. J. Remote Sens. 17(12), 2215–2242 (1996).
10. F. A. Kruse et al., “Mapping alteration minerals at prospect, outcrop and drill core scales
using imaging spectrometry,” Int. J. Remote Sens. 33(6), 1780–1798 (2012).
11. S. Asadzadeh and C. R. de Souza Filho, “A review on spectral processing methods for
geological remote sensing,” Int. J. Appl. Earth Obs. Geoinf. 47, 69–90 (2016).
12. D. Ramakrishnan and R. Bharti, “Hyperspectral remote sensing and geological applica-
tions,” Curr. Sci., Special Section: Hyperspectral Remote Sens. 108(5), 879–891 (2015).
13. E. Bedini, “The use of hyperspectral remote sensing for mineral exploration: a review,”
J. Hyperspectral Remote Sens. 7(4), 189–211 (2017).
14. G. R. Hunt, “Spectral signatures of particulate minerals in the visible and near infrared,”
Geophysics 42(3), 501–513 (1977).
15. R. N. Clark et al., “Imaging spectroscopy: earth and planetary remote sensing with the
USGS Tetracorder and expert systems,” J. Geophys. Res. Planets 108(E12), 1 (2003).
16. D. Coulter et al., “Advances in spectral geology and remote sensing: 2008–2017,” in
Proc. Exploration 17: Sixth Decennial Int. Conf. Mineral Exploration, V. Tschirhart and
M. D. Thomas, Eds., pp. 23–50 (2017).
17. A. B. Pour and M. Hashim, “ASTER, ALI and Hyperion sensors data for lithological
mapping and ore minerals exploration,” Springerplus 3(1), 130 (2014).

18. R. N. Clark et al., “The US Geological Survey, Digital Spectral Reflectance Library:
Version 1: 0.2 to 3.0 microns,” U.S. Geological Survey Open File Report, 93-592 (1993).
19. R. N. Clark, “Spectroscopy of rocks and minerals, and principles of spectroscopy,” in
Remote Sensing for the Earth Sciences: Manual of Remote Sensing, 3rd ed., Vol. 3,
A. N. Rencz, Ed., John Wiley & Sons, Inc., New York (1999).
20. J. W. Salisbury and N. Vergo, “Infrared (2.1–25 um) spectra of minerals,” JCPDS
International Centre for Diffraction Data, PA (1991).
21. J. B. Percival et al., “Customized spectral libraries for effective mineral exploration: mining
National Mineral Collections,” Clays Clay Miner. 66(3), 297–314 (2018).
22. A. F. Goetz, B. Curtiss, and D. A. Shiley, “Rapid gangue mineral concentration measurement
over conveyors by NIR reflectance spectroscopy,” Miner. Eng. 22(5), 490–499 (2009).
23. F. A. Kruse, “Advances in hyperspectral remote sensing for geologic mapping and explo-
ration,” in Proc. 9th Aust. Remote Sens. Conf. (1998).
24. P. Shippert, “Why use hyperspectral imagery?” Photogramm. Eng. Remote Sens. 70(4),
377–396 (2004).
25. K. V. Kale et al., “A research review on hyperspectral data processing and analysis algo-
rithms,” Proc. Natl. Acad. Sci. India Sect. A 87(4), 541–55 (2017).
26. B.-C. Gao et al., “Atmospheric correction algorithms for hyperspectral remote sensing data
of land and ocean,” Remote Sens. Environ. 113, S17–S24 (2009).
27. F. A. Kruse, “Mineral mapping with AVIRIS and EO-1 hyperion” (2004).
28. S. Minu, A. Shetty, and B. Gopal, “Review of preprocessing techniques used in soil prop-
erty prediction from hyperspectral data,” Cogent Geosci. 2(1), 1145878 (2016).
29. D. A. Roberts, Y. Yamaguchi, and R. J. P. Lyon, “Comparison of various techniques for
calibration of AIS data,” JPL Proc. 2nd Airborn Imaging Spectrometer Data Anal.
Workshop, pp. 21–30 (1986).
30. J. E. Conel et al., “AIS-2 radiometry and a comparison of methods for the recovery of
ground reflectance,” Airborne Imaging Spectrometer Data Analysis Workshop (1987).
31. P. N. Reinersman, K. L. Carder, and F.-I. R. Chen, “Satellite-sensor calibration verification
with the cloud-shadow method,” Appl. Opt. 37(24), 5541–9 (1998).
32. L. S. Bernstein et al., “Validation of the quick atmospheric correction (QUAC)
algorithm for VNIR-SWIR multi-and hyperspectral imagery,” Proc. SPIE 5806, 668–678
(2005).
33. A. F. Goetz et al., “Imaging spectrometry for earth remote sensing,” Science 228(4704),
1147–1153 (1985).
34. E. Ben-Dor, B. Kindel, and A. Goetz, “Quality assessment of several methods to recover
surface reflectance using synthetic imaging spectroscopy data,” Remote Sens. Environ.
90(3), 389–404 (2004).
35. B.-C. Gao, K. B. Heidebrecht, and A. F. Goetz, “Derivation of scaled surface reflectances
from AVIRIS data,” Remote Sens. Environ. 44(2-3), 165–178 (1993).
36. R. Richter, “Atmospheric correction of DAIS hyperspectral image data,” Comput. Geosci.
22(7), 785–793 (1996).
37. S. M. Adler-Golden et al., “Atmospheric correction for shortwave spectral imagery based
on MODTRAN4,” Proc. SPIE 3753, 61–69 (1999).
38. K. Staenz, T. Szeredi, and J. Schwarz, “ISDAS—a system for processing/analyzing hyper-
spectral data,” Can. J. Remote Sens. 24(2), 99–113 (1998).
39. Z. Qu, B. C. Kindel, and A. F. Goetz, “The high accuracy atmospheric correction for hyper-
spectral data (HATCH) model,” IEEE Trans. Geosci. Remote Sens. 41(6), 1223–1231
(2003).
40. A. Agarwal et al., “Efficient hierarchical-PCA dimension reduction for hyperspectral
imagery,” in IEEE Int. Symp. Signal Process. and Inf. Technol., IEEE (2007).
41. G. Luo et al., “Minimum noise fraction versus principal component analysis as a prepro-
cessing step for hyperspectral imagery denoising,” Can. J. Remote Sens. 42(2), 106–116
(2016).
42. R. Rajagopal and V. Ranganathan, “Evaluation of effect of unsupervised dimensionality
reduction techniques on automated arrhythmia classification,” Biomed. Signal Process.
Control 34, 1–8 (2017).

43. A. Hyvärinen, “Independent component analysis: recent advances,” Philos. Trans. R. Soc.,
A 371(1984), 20110534 (2013).
44. L. Zhang et al., “Dimensionality reduction based on clonal selection for hyperspectral
imagery,” IEEE Trans. Geosci. Remote Sens. 45(12), 4172–4186 (2007).
45. T. Zhang, D. Tao, and J. Yang, “Discriminative locality alignment,” Lect. Notes Comput.
Sci. 5302, 725–738 (2008).
46. W. Li et al., “Locality-preserving dimensionality reduction and classification for
hyperspectral image analysis,” IEEE Trans. Geosci. Remote Sens. 50(4), 1185–1198
(2012).
47. N. H. Ly, Q. Du, and J. E. Fowler, “Sparse graph-based discriminant analysis for hyper-
spectral imagery,” IEEE Trans. Geosci. Remote Sens. 52(7), 3872–3884 (2014).
48. L. Zhang et al., “Ensemble manifold regularized sparse low-rank approximation for multi-
view feature embedding,” Pattern Recognit. 48(10), 3102–3112 (2015).
49. N. Shental et al., “Adjustment learning and relevant component analysis,” Lect. Notes
Comput. Sci. 2353, 776–790 (2002).
50. J. Goldberger et al., “Neighbourhood components analysis,” in Advances in Neural Inf.
Process. Syst., Vol. 17, pp. 513–20 (2004).
51. J. V. Davis et al., “Information-theoretic metric learning,” in Proc. 24th Int. Conf. Mach.
Learn. (2007).
52. Y. Dong et al., “Dimensionality reduction and classification of hyperspectral images using
ensemble discriminative local metric learning,” IEEE Trans. Geosci. Remote Sens. 55(5),
2509–2524 (2017).
53. S. T. Roweis and L. K. Saul, “Nonlinear dimensionality reduction by locally linear embed-
ding,” Science 290(5500), 2323–2326 (2000).
54. C. Chen et al., “Constrained Laplacian eigenmap for dimensionality reduction,”
Neurocomputing 73(4-6), 951–958 (2010).
55. C. M. Bachmann, T. L. Ainsworth, and R. A. Fusina, “Exploiting manifold geometry in
hyperspectral imagery,” IEEE Trans. Geosci. Remote Sens. 43(3), 441–454 (2005).
56. X. He et al., “Neighborhood preserving embedding,” in Tenth IEEE Int. Conf. Comput.
Vision, Vol. 1, IEEE (2005).
57. H. Huang, M. Chen, and Y. Duan, “Dimensionality reduction of hyperspectral image using
spatial-spectral regularized sparse hypergraph embedding,” Remote Sensing 11(9), 1039
(2019).
58. W. Du et al., “Semi-supervised dimension reduction based on hypergraph embedding for
hyperspectral images,” Int. J. Remote Sens. 39(6), 1696–1712 (2018).
59. Y. Zhou, J. Peng, and C. P. Chen, “Dimension reduction using spatial and spectral regu-
larized local discriminant embedding for hyperspectral image classification,” IEEE Trans.
Geosci. Remote Sens. 53(2), 1082–1095 (2015).
60. F. Kowkabi, H. Ghassemian, and A. Keshavarz, “A fast spatial–spectral preprocessing
module for hyperspectral endmember extraction,” IEEE Geosci. Remote Sens. Lett. 13(6),
782–786 (2016).
61. A. Plaza et al., “A quantitative and comparative analysis of endmember extraction algorithms
from hyperspectral data,” IEEE Trans. Geosci. Remote Sens. 42(3), 650–663 (2004).
62. D. C. Heinz and I. C. Chein, “Fully constrained least squares linear spectral mixture analy-
sis method for material quantification in hyperspectral imagery,” IEEE Trans. Geosci.
Remote Sens. 39(3), 529–545 (2001).
63. M. Parente and A. Plaza, “Survey of geometric and statistical unmixing algorithms for
hyperspectral images,” in 2nd Workshop Hyperspectral Image and Signal Process.:
Evol. Remote Sens., IEEE (2010).
64. K. Wu et al., “A novel endmember extraction method using sparse component analysis for
hyperspectral remote sensing imagery,” IEEE Access 6, 75206–75215 (2018).
65. T. Nouri, M. M. Oskouei, and H. Zekri, “A comparison study of ORASIS and VCA
for mineralogical unmixing of hyperspectral data,” J. Indian Soc. Remote Sens. 44(5),
723–733 (2016).
66. C.-I. Chang, Hyperspectral Data Exploitation: Theory and Applications, John Wiley &
Sons, Hoboken, New Jersey (2007).

67. L. Miao and H. Qi, “Endmember extraction from highly mixed data using minimum
volume constrained nonnegative matrix factorization,” IEEE Trans. Geosci. Remote Sens.
45(3), 765 (2007).
68. T. Nouri et al., “Improvement of the MVC-NMF problem using particle swarm optimi-
zation for mineralogical unmixing of noisy hyperspectral data,” J. Indian Soc. Remote
Sens. 47(4), 541–550 (2019).
69. J. Li and J. M. Bioucas-Dias, “Minimum volume simplex analysis: a fast algorithm to
unmix hyperspectral data,” in IEEE Int. Geosci. and Remote Sens. Symp. (2008).
70. J. Li et al., “Minimum volume simplex analysis: a fast algorithm for linear hyperspectral
unmixing,” IEEE Trans. Geosci. Remote Sens. 53(9), 5067–5082 (2015).
71. T.-H. Chan et al., “A convex analysis-based minimum-volume enclosing simplex algorithm
for hyperspectral unmixing,” IEEE Trans. Signal Process. 57(11), 4418–4432 (2009).
72. M. D. Craig, “Minimum-volume transforms for remotely sensed data,” IEEE Trans.
Geosci. Remote Sens. 32(3), 542–552 (1994).
73. J. Broadwater, A. Banerjee, and P. Burlina, “Kernel methods for unmixing hyperspectral
imagery,” in Kernel Methods for Remote Sensing Data Analysis, pp. 249–70 (2009).
74. R. Heylen, D. Burazerovic, and P. Scheunders, “Non-linear spectral unmixing by geodesic
simplex volume maximization,” IEEE J. Sel. Top. Signal Process. 5(3), 534–542 (2011).
75. M. T. Eismann and R. C. Hardie, “Stochastic spectral unmixing with enhanced endmember
class separation,” Appl. Opt. 43(36), 6596–608 (2004).
76. D. Lu and Q. Weng, “A survey of image classification methods and techniques for improv-
ing classification performance,” Int. J. Remote Sens. 28(5), 823–870 (2007).
77. M. Pal, T. Rasmussen, and A. Porwal, “Optimized lithological mapping from multispectral
and hyperspectral remote sensing images using fused multi-classifiers,” Remote Sensing
12(1), 177 (2020).
78. D. Chutia et al., “Hyperspectral remote sensing classifications: a perspective survey,”
Trans. GIS 20(4), 463–490 (2016).
79. S. Amini, S. Homayouni, and A. Safari, “Semi-supervised classification of hyperspectral
image using random forest algorithm,” in IEEE Geosci. and Remote Sens. Symp., IEEE (2014).
80. R. Heylen, M. Parente, and P. Gader, “A review of nonlinear hyperspectral unmixing
methods,” IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 7(6), 1844–1868 (2014).
81. S. S. Nath et al., “A survey of image classification methods and techniques,” in Int. Conf.
Control, Instrum., Commun. and Comput. Technol., IEEE (2014).
82. J. F. Mustard and J. M. Sunshine, “Spectral analysis for earth science: investigations using
remote sensing data,” in Remote Sensing for the Earth Sciences: Manual of Remote
Sensing, A. N. Rencz, Ed., Vol. 3, pp. 251–307, John Wiley & Sons, Inc, New York (1999).
83. F. van der Meer, “Analysis of spectral absorption features in hyperspectral imagery,” Int. J.
Appl. Earth Obs. Geoinf. 5(1), 55–68 (2004).
84. J. A. Richards and X. Jia, “Interpretation of hyperspectral image data,” in Remote Sensing
Digital Image Analysis: An Introduction, J. D. Richards and X. Jia, Eds., pp. 359–88,
Springer, Berlin, Heidelberg(2006).
85. C. Kumar et al., “Automated lithological mapping by integrating spectral enhancement
techniques and machine learning algorithms using AVIRIS-NG hyperspectral data in
Gold-bearing granite-greenstone rocks in Hutti, India,” Int. J. Appl. Earth Obs. Geoinf.
86, 102006 (2020).
86. S. Salehi, C. Mielke, and C. Rogass, “Mapping ultramafic complexes using airborne im-
aging spectroscopy and spaceborne data in Arctic regions with abundant lichen cover,
a case study from the Niaqornarssuit complex in South West Greenland,” Eur. J. Remote
Sens. 53(1), 156–175 (2020).
87. J. Harris et al., “Mapping lithology in Canada’s Arctic: application of hyperspectral data
using the minimum noise fraction transformation and matched filtering,” Can. J. Earth Sci.
42(12), 2173–2193 (2005).
88. X. Zhang and P. Li, “Lithological mapping from hyperspectral data by improved use of
spectral angle mapper,” Int. J. Appl. Earth Obs. Geoinf. 31, 95–109 (2014).
89. A. Dadon et al., “Examination of spaceborne imaging spectroscopy data utility for strati-
graphic and lithologic mapping,” J. Appl. Remote Sens. 5(1), 053507 (2011).

90. J. F. Huntington, “The role of remote sensing in finding hydrothermal mineral deposits on
earth,” Ciba Found Symp. 202, 214–31 (1996).
91. R. Sillitoe, “Porphyry copper systems,” Econ. Geol. 105, 3–41 (2010).
92. X. Zhou et al., “Multi-scale integrated application of spectral geology and remote sensing
for mineral exploration,” in Proc. 6th Decennial Int. Conf. Mineral Exploration, Toronto,
ON (2017).
93. F. Pirajno, “A classification of mineral systems, overviews of plate tectonic margins and
examples of ore deposits associated with convergent margins,” Gondwana Res. 33, 44–62
(2016).
94. L. Liu et al., “Mapping alteration using imagery from the Tiangong-1 hyperspectral space-
borne system: example for the Jintanzi gold province, China,” Int. J. Appl. Earth Obs.
Geoinf. 59, 31–41 (2017).
95. D. A. John et al., “Descriptive models for epithermal gold-silver deposits: chapter Q in
mineral deposit models for resource assessment,” Report No.: 2328-0328, US Geological
Survey (2018).
96. A. Crosta et al., “Targeting key alteration minerals in epithermal deposits in Patagonia,
Argentina, using ASTER imagery and principal component analysis,” Int. J. Remote
Sens. 24(21), 4233–40 (2003).
97. F. van der Meer et al., “Wavelength feature mapping as a proxy to mineral chemistry for
investigating geologic systems: an example from the Rodalquilar epithermal system,” Int.
J. Appl. Earth Obs. Geoinf. 64, 237–248 (2018).
98. O. Corumluoglu, A. Vural, and I. Asri, “Determination of Kula basalts (geosite) in Turkey
using remote sensing techniques,” Arabian J. Geosci. 8(11), 10105–10117 (2015).
99. A. Vural, Ö. Corumluoglu, and I. Asri, “Remote sensing technique for capturing and explo-
ration of mineral deposit sites in Gumushane metallogenic province, NE Turkey,” J. Geol.
Soc. India 90(5), 628–633 (2017).
100. R. A. Ayuso et al., “Porphyry copper deposit model: chapter B in mineral deposit models
for resource assessment,” Report No.: 2328-0328, US Geological Survey (2010).
101. S. Halley, J. Dilles, and R. Tosdal, “Footprints: the hydrothermal alteration and geochemi-
cal dispersion around porphyry copper deposits,” Society of Economic Geologists
Newsletter 100(1), 12–17 (2015).
102. C. A. Bishop, J. G. Liu, and P. J. Mason, “Hyperspectral remote sensing for mineral
exploration in Pulang, Yunnan province, China,” Int. J. Remote Sens. 32(9), 2409–2426
(2011).
103. G. E. Graham et al., “Application of imaging spectroscopy for mineral exploration in
Alaska: a study over porphyry Cu deposits in the eastern Alaska Range,” Econ. Geol.
113(2), 489–510 (2018).
104. M. H. Zadeh et al., “Sub-pixel mineral mapping of a porphyry copper belt using EO-1
Hyperion data,” Adv. Space Res. 53(3), 440–451 (2014).
105. M. D. Barton, Iron Oxide (-Cu-Au-REE-P-Ag-U-Co) Systems. Treatise on Geochemistry,
2nd ed., pp. 515–41, Elsevier Inc, San Diego (2013).
106. L. Corriveau et al., “Alteration vectoring to IOCG (U) deposits in frontier volcano-plutonic
terrains, Canada,” in Proc. Exploration (2007).
107. M. C. Tappert et al., “The mineral chemistry, near-infrared, and mid-infrared reflectance
spectroscopy of phengite from the Olympic Dam IOCG deposit, South Australia,” Ore
Geol. Rev. 53, 26–38 (2013).
108. C. Laukamp et al., “Hydrothermal mineral alteration patterns in the Mount Isa Inlier
revealed by airborne hyperspectral data,” Aust. J. Earth Sci. 58(8), 917–936 (2011).
109. W. P. Shanks, III et al., “Volcanogenic massive sulfide occurrence model: chapter C in
mineral deposit models for resource assessment,” Report No.: 2328-0328, US Geological
Survey (2012).
110. A.-L. Bonnet and L. Corriveau, “Alteration vectors to metamorphosed hydrothermal sys-
tems in gneissic terranes,” in Mineral Deposits of Canada: A Synthesis of Major Deposit-
Types, District Metallogeny, the Evolution of Geological Provinces, and Exploration
Methods, W. D. Goodfellow, Eds., Vol. 5, pp. 1035–49, Geological Association of
Canada, Mineral Deposits Division, Special Publication (2007).

111. F. J. van Ruitenbeek et al., “Characterization of the hydrothermal systems associated

with Archean VMS-mineralization at Panorama, Western Australia, using hyperspectral,
geochemical and geothermometric data,” Ore Geol. Rev. 45, 33–46 (2012).
112. P. Samani, S. Prizomwala, and A. Rajawat, “Assessing the mineral alteration in Ambaji–
Deri Region (Northwestern India) using hyperspectral remote sensing,” J. Indian Soc.
Remote Sens. 49, 249–257 (2020).
113. K. Laakso et al., “Application of airborne, laboratory, and field hyperspectral methods to
mineral exploration in the Canadian Arctic: recognition and characterization of volcano-
genic massive sulfide-associated hydrothermal alteration in the Izok Lake deposit area,
Nunavut, Canada,” Econ. Geol. 110(4), 925–941 (2015).
114. L. D. Meinert, “Skarns and skarn deposits,” Geosci. Canada 19(4), 145–162 (1992).
115. Y. Xu, J. Chen, and P. Meng, “Detection of alteration zones using hyperspectral remote
sensing data from Dapingliang skarn copper deposit and its surrounding area, Shanshan
county, Xinjiang Uygur autonomous region, China,” J. Vis. Commun. Image Represent.
58, 67–78 (2019).
116. J. Tian et al., “Short wavelength infra-red (SWIR) characteristics of hydrothermal alter-
ation minerals in skarn deposits: example from the Jiguanzui Cu–Au deposit, Eastern
China,” Ore Geol. Rev. 106, 134–149 (2019).
117. B. Dkhala et al., “Hyperspectral field spectroscopy and Sentinel-2 multispectral data for
minerals with high pollution potential content estimation and mapping,” Sci. Total
Environ. 740, 140160 (2020).
118. S. Abbasi et al., “Bioaccessibility of barium from barite contaminated soils based on
gastric phase in vitro data and plant uptake,” Chemosphere 144, 1421–1427 (2016).
119. X. Yang et al., “Use of hyperspectral imagery to detect affected vegetation and heavy metal
polluted areas: a coal mining area, China,” Geocarto Int. 1–20 (2020).
120. A. Riaza et al., “Mapping acidic water in an acid mine drainage-contaminated river with
hyperspectral airborne hymap data,” Odiel River, SW Spain (2012).
121. C. Ehrler, C. Fischer, and M. Bachmann, “Hyperspectral Remote Sensing Applications in
Mining Impact Analysis,” in 34th Int. Symp. Remote Sens. Environ. (2011).
122. E. Choe et al., “Mapping of heavy metal pollution in stream sediments using combined
geochemistry, field spectroscopy, and hyperspectral remote sensing: a case study of the
Rodalquilar mining area, SE Spain,” Remote Sens. Environ. 112(7), 3222–3233 (2008).
123. G. A. Swayze et al., “Using imaging spectroscopy to map acidic mine waste,” Environ. Sci.
Technol. 34(1), 47–54 (2000).
124. S. Raval, R. Merton, and D. Laurence, “Mine tailings water mapping using CHRIS Proba
imagery,” in 4th Workshop Hyperspectral Image and Signal Process.: Evol. Remote Sens.,
IEEE (2012).
125. C. Mielke et al., “Spaceborne mine waste mineralogy monitoring in South Africa, appli-
cations for modern push-broom missions: hyperion/OLI and EnMAP/Sentinel-2,” Remote
Sens. 6(8), 6790–6816 (2014).
126. G. E. Davies and W. M. Calvin, “Mapping acidic mine waste with seasonal airborne hyper-
spectral imagery at varying spatial scales,” Environ. Earth Sci. 76, 432 (2017).
127. B. E. Teledyne, “DESIS: frequently asked questions,” https://tbecom/geospatial/faq-DESIS.
Sima Peyghambari received her PhD in geology (petrology) from Shahid Bahonar University
of Kerman. She is an assistant professor in Geology Department of Payame-Noor University of
Tehran. She is Master of Science student in Engineering under supervising of Dr. Zhang at
University of New Brunswick. She has authored or coauthored over 10 research papers.
Yun Zhang received his PhD from the Free University of Berlin. He is a fellow of Canadian
Academy of Engineering (FCAE), and a professor in remote sensing at the University of
New Brunswick, Canada. He has authored or coauthored over 260 research papers. The research
outcomes from his lab have resulted 15 patents, 3 new patent applications, and more than ten
technology licenses. In addition, he received many prestigious national and international
research awards.

Terms of Use:View
https://www.spiedigitallibrary.org/terms-of-use
publication stats

Hyperspectral Remote Sensing in Lithological Mapping, Mineral Exploration, and Environmental Geology: An Updated Review

Uploaded by

Copyright:

Available Formats

Hyperspectral Remote Sensing in Lithological Mapping, Mineral Exploration, and Environmental Geology: An Updated Review

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Hyperspectral Remote Sensing in Lithological Mapping, Mineral Exploration, and Environmental Geology: An Updated Review

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Hyperspectral remote sensing in lithological mapping, mineral exploration, and

Article in Journal of Applied Remote Sensing · July 2021

Sima Peighambari Yun Zhang

SEE PROFILE SEE PROFILE

The user has requested enhancement of the downloaded file.

Sima Peyghambari* and Yun Zhang

*Address all correspondence to Sima Peyghambari, sima.peyghambari@unb.ca

Journal of Applied Remote Sensing 031501-1 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

3 Spectral Properties of Different Minerals

Journal of Applied Remote Sensing 031501-2 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

Satellites/ Launch Number of Spectral Spectral GSD Swath Spectral imaging

Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

Journal of Applied Remote Sensing

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

Journal of Applied Remote Sensing 031501-4 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

4 Hyperspectral Data Processing

4.1 Hyperspectral Image Preprocessing

Journal of Applied Remote Sensing 031501-5 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

4.2 Dimensionality Processing

4.3 Endmember Extraction Methods

Journal of Applied Remote Sensing 031501-6 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

5 Image Classification Techniques

Journal of Applied Remote Sensing 031501-7 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

Journal of Applied Remote Sensing 031501-8 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

Journal of Applied Remote Sensing 031501-9 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

6.1 Lithological Mapping

Journal of Applied Remote Sensing 031501-10 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

6.2 Mineral Mapping and Mineral Exploration

Journal of Applied Remote Sensing 031501-11 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

Fig. 5 Generalized alteration-mineralization zoning pattern for PCDs.91

Journal of Applied Remote Sensing 031501-12 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

Fig. 6 Common alteration minerals in different ore deposits.92

Journal of Applied Remote Sensing 031501-13 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

Journal of Applied Remote Sensing 031501-14 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

Journal of Applied Remote Sensing 031501-15 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

Journal of Applied Remote Sensing 031501-16 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

Journal of Applied Remote Sensing 031501-17 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

6.3 Environmental Geology

Journal of Applied Remote Sensing 031501-18 Jul–Sep 2021 • Vol. 15(3)

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Applied-Remote-Sensing on 14 Jul 2021

7 Discussion and Conclusion

Journal of Applied Remote Sensing 031501-19 Jul–Sep 2021 • Vol. 15(3)